567962 research-article2015
JIVXXX10.1177/0886260514567962Journal of Interpersonal ViolenceScholes-Balog et al.
Article
Relationships Between Sport Participation, Problem Alcohol Use, and Violence: A Longitudinal Study of Young Adults in Australia
Journal of Interpersonal Violence 2016, Vol. 31(8) 1501–1530 © The Author(s) 2015 Reprints and permissions: sagepub.com/journalsPermissions.nav DOI: 10.1177/0886260514567962 jiv.sagepub.com
Kirsty E. Scholes-Balog, PhD,1 Sheryl A. Hemphill, PhD,1,2,3,4 Peter J. Kremer, PhD,3 and John W. Toumbourou, PhD3,4
Abstract There is a growing body of evidence suggesting a link between sport participation and violent behavior outside of the sporting context. However, there have been few studies that have investigated the basis of this relationship. The current study examined longitudinal relationships between sport participation, problem alcohol use, and various violent behaviors, and whether sport participation moderates relationships between problem alcohol use and violence. The sample comprised 2,262 young adults (55% female, age range at Time 1 = 17-24 years) from Victoria, Australia, surveyed in 2010 and 2012. When controlling for common risk factors, substance use, and past violence, sport participation was not associated with any violent behaviors 2 years later. However, sport participation moderated the relationship between problem alcohol use and fighting, whereby problem
1Australian
Catholic University, Fitzroy, Victoria, Australia University of Melbourne, Parkville, Victoria, Australia 3Deakin University, Geelong, Victoria, Australia 4Murdoch Childrens Research Institute, Parkville, Victoria, Australia 2The
Corresponding Author: Kirsty E. Scholes-Balog, School of Psychology, Australian Catholic University, 115 Victoria Parade, Fitzroy, Victoria 3065, Australia. Email: [email protected]
Downloaded from jiv.sagepub.com at Australian National University on June 4, 2016
1502
Journal of Interpersonal Violence 31(8)
alcohol use was associated with engaging in fights 2 years later for sport participants, but not for nonparticipants. These findings suggest that it is not sport participation per se that influences later violence but the drinking norms or culture embedded within certain sporting contexts. Prevention approaches that address the drinking culture and social approval of excessive alcohol consumption within sporting contexts may reduce the incidence of violent behavior in the community. Keywords sport, alcohol use problems, violence, young adult Media reports of high profile athletes’ involvement in aggressive and violent incidents have focused attention on the relationship between sport participation and violence. While many of these violent or aggressive incidents take place on the field in the context of sports competition, there are also reports of violence against members of the community, and often with respect to women in intimate and nonintimate situations. Despite this anecdotal evidence of a link between sport involvement and community violence, there have been relatively few empirical studies that have been able to appropriately examine this relationship (Kimble, Russo, Bergman, & Galindo, 2010). Of the studies that do exist, many have found evidence to suggest that involvement in sport may be associated with violence and aggression outside of the sporting context. Participation in sport has been shown to be associated with family violence (Bloom & Smith, 1996), laboratory measures of aggression (D. B. Huang, Cherek, & Lane, 1999), sexual violence/aggression (Frinter & Rubinson, 1993; Koss & Gaines, 1993; Murnen & Kohlman, 2007), violent delinquency (Segrave, Moreau, & Hastad, 1985), and physical and/or verbal aggression/violence (Burton & Marshall, 2005; Endresen & Oleweus, 2005; Linville & Huebner, 2005; Nixon, 1997; O’Brien et al., 2012). Gender (Linville & Huebner, 2005; Miller, Melnick, Farrell, Sabo, & Barnes, 2006; Nixon, 1997; O’Brien et al., 2012), type of sport (e.g., contact or noncontact sport; Nixon, 1997), level of competitiveness (Bloom & Smith, 1996), and whether or not an individual perceives himself or herself as an athlete or “jock” (Miller et al., 2006) may also influence these relationships. It should be noted that not all studies have observed relationships between sport participation and violent behavior off-field (Kimble et al., 2010). For example, among college men, participation in sport (membership on a competitive sports team) was not found to be predictive of perpetration of verbal, physical, or sexual violence 3 months later when prior violent behavior was taken into account (Gidycz, Warkentin, & Orchowski, 2007).
Downloaded from jiv.sagepub.com at Australian National University on June 4, 2016
Scholes-Balog et al.
1503
There are a number of theories that help explain potential relationships between sport participation and violence. Social learning theory suggests that aggressive behaviors are learned through the confrontational nature of some sports (e.g., tackling, use of physical force, fighting skills in sports such as boxing or wrestling; Endresen & Oleweus, 2005). Furthermore, enactive learning and contact with aggressive role models during training and competition may enhance aggressive behaviors in the sporting context and the normalization or social approval of such behaviors (Endresen & Oleweus, 2005). Cultural spillover theory suggests that socially approved forms of aggression or violence can then have a carryover effect to more forbidden forms of aggression or violence, such as outside of the sporting context (Bloom & Smith, 1996). Participation in sport has also been consistently linked with alcohol use, with many studies showing that members of sporting clubs (Poortinga, 2007) and athletes (Gmel, Kuendig, & Daeppen, 2009; Leichliter, Meilman, Presley, & Cashin, 1998; O’Brien et al., 2012; O’Brien, Kolt, Webber, & Hunter, 2010) report greater alcohol consumption and more harm associated with alcohol consumption than nonsports people. For example, a large study of college students in the United States found that athletes reported more heavy episodic drinking, heavier alcohol use, and a greater level of drinking-related harms than nonathletes (Nelson & Wechsler, 2001). Furthermore, level of involvement in sport may influence the relationship between sport participation and alcohol use, with elite sports people reporting higher rates of hazardous drinking than both nonelite sports people and nonsports people (O’Brien, Ali, Cotter, O’Shea, & Stannard, 2007; O’Brien, Blackie, & Hunter, 2005; Sonderlund et al., 2014). There is also a well-established link between alcohol use and violent behavior. Experimental studies have shown that alcohol has an indirect causal influence on violent behavior, by way of cognitive, emotional, or physiological changes caused by alcohol consumption, which increase the likelihood of aggressive responses (for a review, see Exum, 2006). Longitudinally, alcohol use has also been found to predict later violent behavior (outside of the period of intoxication) in a number of studies (Dembo et al., 1993; Ellickson, Tucker, & Klein, 2003; Newcomb & McGee, 1989; Scholes-Balog, Hemphill, Kremer, & Toumbourou, 2013), while other studies have suggested bidirectional or reciprocal longitudinal relationships between alcohol use and violence (Scholes-Balog et al., 2013; White, Loeber, Stouthamer-Loeber, & Farrington, 1999; Windle, 1990; Zue, Zimmerman, & Cunningham, 2009). Longitudinal studies also support an indirect causal relationship between alcohol use and later violent behavior, with factors such as affiliation with antisocial and substance using peers appearing to explain these relationships (e.g., Scholes-Balog et al., 2013).
Downloaded from jiv.sagepub.com at Australian National University on June 4, 2016
1504
Journal of Interpersonal Violence 31(8)
While the bulk of the literature has examined violence in relation to sports participation and alcohol consumption separately, there is a small body of literature that has attempted to evaluate the relationship among all three factors (for a review, see Sonderlund et al., 2014). A qualitative study among athletes found that athletes often attribute alcohol use as the cause of their violent behavior outside of the sporting context (Pappas, McKenry, & Skilken Catlett, 2004). Similarly, off-field alcohol-related aggression and antisocial behavior was found to be common among professional Australian Football League players (Dietze, Fitzgerald, & Jenkinson, 2008). In this study, 26% of players surveyed reported getting involved in a fight (either verbal or physical) while drinking (Dietze et al., 2008). Likewise, a study of Australian university students found that university sport participants were more likely than nonsport participants to have displayed aggressive behavior (insulted or assaulted someone) when intoxicated in the past year (O’Brien et al., 2012). The university sport participant sample included participants in both individual (e.g., running, squash) and team sports (e.g., football, cricket). In this study, gender also played a role, with alcoholrelated violence and aggression more likely among male university sport participants than among female sport participants (O’Brien et al., 2012). Given the mounting evidence suggesting a higher rate of alcohol consumption and violence among athletes, relative to nonathletes, Sonderlund et al. (2014) suggested that sport participation may have a moderating effect on the positive relationship between alcohol and violence. However, this suggestion has yet to be empirically tested. Nonetheless, the context in which drinking takes place (and the associated dynamics of such context) has been shown to play a role in alcohol-related violence. For instance, public drinking (as compared with private drinking) is a strong predictor of violence (Rossow, 1996; Tomsen, 1997). Similarly, drinking among sporting spectators is thought to play a role in spectator aggression and violence at sporting matches (Bormann & Stone, 2001; Moore, Shepherd, Eden, & Sivarajasingam, 2007), with the social dynamics of the spectator crowd thought to also drive this relationship (Roberts & Benjamin, 2000). It should be noted that the bulk of the literature examining alcohol use, sport participation, and violence is cross-sectional in nature. One longitudinal study found that neither alcohol use nor sport participation predicted violence or aggression, when controlling for past history of violence and aggression (Gidycz et al., 2007). Another longitudinal study did not find sport participation itself to be predictive of violence, but rather “jock” identity (those who subjectively perceive themselves, and are perceived by others, as athletes or “jocks”) was associated with a higher frequency of violence (Miller et al., 2006). Furthermore, “jock” identity moderated the relationship between
Downloaded from jiv.sagepub.com at Australian National University on June 4, 2016
Scholes-Balog et al.
1505
alcohol use and violence, whereby alcohol use predicted violence only among nonjocks (Miller et al., 2006). There are a number of limitations of the published literature to date. First, the literature is dominated by studies of North American college samples (Sonderlund et al., 2014). Factors such as drinking cultures, sports team dynamics and subcultures, and social norms may vary between countries. For example, in Australia there is no named subculture synonymous with “jock,” though such a subculture may exist without name, or simply less pejoratively termed “sporty” or “athlete.” Nonetheless, literature investigating “jock” identity is not easily comparable in an Australian setting. Similarly, in countries where sport participation is not as valued or esteemed as in the United States, the customs associated with these sports may be less likely to influence participants’ behavior (Sonderlund et al., 2014). Furthermore, as mentioned previously, the vast majority of published studies investigating sport participation, alcohol use, and violence are of cross-sectional design and therefore are unable to assess temporal ordering or control for prior violent behavior when assessing relationships between sport participation and violence. In addition, most studies fail to control for other factors that may explain or account for any observed relationships between sport participation, alcohol use, and violence, such as use of illicit substances and affiliation with deviant peers (Herrenkohl et al., 2007; McMorris, Catalano, Kim, Toumbourou, & Hemphill, 2011; Scholes-Balog et al., 2013). Given the lack of research directed specifically at understanding the relationships between sport participation, alcohol use, and violence, together with the limitations outlined above, the present study intended to investigate longitudinal relationships between sport participation, problem alcohol use, and violence among a community-based Australian sample. Specifically, this study examines relationships with four violent behaviors: physical intimate partner violence (IPV), emotional IPV, fighting, and assault. It was hypothesized that (a) both alcohol use problems and frequency of sport participation will predict violent behavior 2 years later; (b) as suggested by Sonderlund et al. (2014), sport participation will moderate the relationship between alcohol use problems and violence; and (c) gender will moderate the relationship between alcohol use, sport participation, and violence.
Method Participants Young adults from Victoria, Australia, who were involved in the International Youth Development Study (IYDS) comprised the sample for this study. In
Downloaded from jiv.sagepub.com at Australian National University on June 4, 2016
1506
Journal of Interpersonal Violence 31(8)
2002, 3,948 Victorian students and their parents were approached to take part in this study. Of these, 2,927 parents provided informed consent (74%). Among those students who had parental informed consent, 2,884 (99%) were recruited into the IYDS and completed the survey. Original sampling for the IYDS has been described in detail elsewhere (McMorris, Hemphill, Toumbourou, Catalano, & Patton, 2007) and was designed to yield a staterepresentative sample of students in Grades 5, 7, and 9 (equivalent to average ages 11, 13, and 15 years). Briefly, the IYDS used a two-stage cluster sampling approach: (a) random selection of public and private schools in Victoria, Australia, stratified according to geographic location, using a probability proportionate to grade-level size sample procedure and (b) random selection of one class at each target grade level (Grades 5, 7, and 9 in 2002) within each school. Longitudinal surveying for the IYDS has occurred between the years 2002 and 2012. The sample for the current study comprised 2,262 (1,260 female, 1,000 male) young adults who were surveyed in both 2010 and 2012 as part of the IYDS longitudinal study and had data available for the measures of violence (this constituted 77% of the original sample who had entered the study in 2002). The current study utilizes these two time points as only questions related to the focus of this article were asked during these surveys. In 2010 (referred to as T1), participants ranged in age from 17 to 24 years of age (M = 21.00, SD = 1.67). In 2012 (referred to as T2), participants ranged in age from 19 to 26 years of age (M = 23.02, SD = 1.66).
Procedure Ethics approval was obtained from The University of Melbourne Human Ethics in Research Committee. Participants were contacted by mail, email, and/or phone and asked to complete the survey online after providing informed consent. Survey completion took approximately 60 min. After completion of the survey, participants received a small thank you gift (gift voucher).
Measures Survey items were drawn from a modified version of the Communities That Care youth survey (Glaser, Lee Van Horn, Arthur, Hawkins, & Catalano, 2005; Hemphill et al., 2011) that was adapted for use in young adults. Sport participation. Participation in sport was assessed at T1 with a question asking, “How many times in the past 12 months have you participated in organized sporting activities (organized by a club, association, or other type
Downloaded from jiv.sagepub.com at Australian National University on June 4, 2016
Scholes-Balog et al.
1507
of organization)?” This item was scored on an 8-point scale from never (1) to 40+ times (8). As this variable was severely skewed, and could not be rectified with transformation, it was dichotomized for analysis: sport participant (1) was defined as having engaged in sport at least once in the past 12 months, sport nonparticipant (0) was defined as not having engaged in sport at all in the past 12 months. Alcohol use problems. The Alcohol Use Disorders Identification Test (AUDIT; Babor, de la Fuente, Saunders, & Grant, 1992) was used to measure alcohol use problems at T1. The AUDIT consists of 10 items, each scored on a 5-point scale. Responses to each item are summed to obtain an overall score that ranges between 0 and 40. Higher scores are indicative of greater severity of alcohol-related problems, with scores more than 8 indicating a strong likelihood of harmful alcohol consumption (Babor et al., 1992). The AUDIT has been shown to be valid and reliable among college samples (e.g., Fleming, Barry, & Macdonald, 1991). Fighting. Engaging in fights was assessed at both T1 and T2. At T1, the item asked, “Have you ever gotten into physical fights with other people?” and the response options were “no,” “yes, but not in the past 12 months,” “yes, once in the past 12 months,” and “yes, more than once in the past 12 months.” At T2, the item asked, “How many times in the past year (12 months) have you got into physical fights with other people?” and the response options ranged from “never” to “10 or more times.” Given the different response options at each time point, for both T1 and T2, these items were dichotomized; a score of 1 indicated that the participant had engaged in fights at least once in the past 12 months, whereas a score of 0 indicated that the participant did not engage in fighting in the past 12 months. Assault. Perpetration of assault was assessed at both T1 and T2. At T1, the item asked, “Have you ever beat up someone so badly that they probably needed a doctor?” and had the same response options as the T1 fighting item. At T2, the item asked “How many times in the past year (12 months) have you beat up someone so badly that they probably needed to see a doctor?” and had the same response options as the T2 fighting item. As above, at T1 and T2 these items were dichotomized; a score of 1 indicated that the participant had perpetrated assault at least once in the past 12 months, whereas a score of 0 indicated that the participant had not perpetrated assault in the past 12 months. IPV. Perpetration of IPV in the past 12 months was assessed at both T1 and T2 with a scale containing five items, each assessing perpetration of different
Downloaded from jiv.sagepub.com at Australian National University on June 4, 2016
1508
Journal of Interpersonal Violence 31(8)
types of behaviors: (a) “I pushed, grabbed, slapped, shoved my partner, or threw something at my partner that could hurt”; (b) “I threatened to end our relationship or leave”; (c) “I threatened to hit or throw something at my partner”; (d) “My partner had a physical injury, sprain, bruise, or small cut because of a fight with me”; and (e) “I insulted, swore, or yelled at my partner.” The IPV questions were only asked of participants who indicated they had been/were currently in a relationship in the past 12 months (n = 1,545 at T1 and n = 1,567 at T2). These items were drawn from the revised Conflict Tactics Scale (Straus, Hamby, Boney-McCoy, & Sugarman, 1996), and in some cases, were modified for the purpose of this study. The frequency of each behavior was scored with response options “this has never happened to me” (0), “not in the past year, but it did happen before” (1), “once in the past year” (2), “twice in the past year” (3), “3-5 times in the past year” (4), “6-10 times in the past year” (5), “11-20 times in the past year” (6) and “more than 20 times in the past year” (7). Two measures of IPV were derived from these items: physical IPV (Items a, c, and d) and emotional IPV (Items b and e). To be consistent with the other measures of violence, these measures were dichotomized; a score of 1 indicated that the participant had perpetrated at least one IPV behavior at least once in the past 12 months, whereas a score of 0 indicated that the participant had not perpetrated IPV in the past 12 months. Common risk factors. A range of risk factors common to both alcohol use and violence were measured at T1 and controlled for in the analyses. Antisocial peers was assessed with a scale containing seven items asking about the number of peers that have engaged in certain antisocial behaviors in the past 12 months (e.g., “In the past year, how many of your best friends have stolen something worth more than $50AUD?”). Each item was scored on a 5-point scale, ranging from none of my friends (0) to 4 of my friends (4). Peer drug use was assessed with a scale containing seven items asking about the number of friends who had used certain types of drugs in the past 12 months (e.g., “In the past year, how many of your best friends have gotten drunk?”). This item was scored on the same rating scale as above. Finally, peer attachment was assessed with two items (e.g., “Think about your four best friends (the friends you feel closest to), in the past year, how many of your best friends have you shared your thoughts and feelings with?”). These items were rated on a 5-point scale from none of my friends (1) to 4 of my friends (5). For each of these scales, scale scores were obtained by taking the mean of the item responses. In addition, use of other substances were included as covariates in the analyses. Cigarette use was assessed with the item “Have you smoked cigarettes in the past year?” This item was scored on a 5-point scale from never
Downloaded from jiv.sagepub.com at Australian National University on June 4, 2016
Scholes-Balog et al.
1509
(1) to almost every day or every day (5). Cannabis use was assessed with the item “In the past year, on how many occasions, if any, have you used marijuana (pot, weed, grass)?” This item was scored on an 8-point scale ranging from never (1) to 40+ times (8). Finally, use of other illicit substances was assessed with eight items asking about the frequency of use of different types of illicit substances (lysergic acid diethylamide [LSD], cocaine, inhalants, stimulants, ecstasy, heroin, other illicit drugs, over the counter drugs for recreational/nonmedical purposes). Each item assessing use of each substance was rated on an 8-point scale ranging from never (1) to 40+ times (8). A composite measure of frequency of other illicit substance use was computed by taking the mean of the eight items. Finally, in addition to age and gender, two demographic variables were included as covariates in the regression analyses. Education was assessed with the question, “What was the highest year level at secondary school you completed?” This item was measured on a 5-point scale, ranging from year 8 or below (5) to year 12 or equivalent (1). Income was included as a proxy measure of socioeconomic status. Income was assessed with the question, “What is your usual, take-home, weekly income from all sources of support (this means after tax is taken out)?” This item was scored on a 10-point scale ranging from $0 (1) to $501 or more (10). Descriptive statistics for the measures can be found in Table 1.
Analyses Data were analyzed using the statistical software STATA IC for Windows (version 13; StataCorp, 2013). The analyses consisted of a number of steps. First, attrition was examined. Of those who were originally recruited into the IYDS in 2002, but were not included in the current study due to drop out, 63% (n = 381) were male. Those included in the current sample also had higher scores at the first wave of data collection (collected in 2002 when participants were aged 10-15 years) on measures of frequency of violence (beat someone up) and antisocial behavior (school suspension, stealing, arrest, carrying weapon, attacking someone) than those who did not complete both the young adult waves of the survey (collected in 2010 and 2012). However, there was no difference on measures of lifetime alcohol use measured in 2002 between those who completed the young adult waves and those who did not. Second, the variance inflation factor (VIF) of each independent variable in each analysis was examined for potential multicollinearity. All VIFs were less than 2.06, indicating no multicollinearity. In Step 3, partially adjusted logistic regression analyses were performed to test relationships between
Downloaded from jiv.sagepub.com at Australian National University on June 4, 2016
1510
Journal of Interpersonal Violence 31(8)
Table 1. Descriptive Statistics for Alcohol Use and Covariate Measures.
T1 AUDIT—Alcohol use problems T2 AUDIT—Alcohol use problems T1 Cigarette use T1 Cannabis use T1 Other illicit drug use T1 Antisocial peers T1 Friends’ drug use T1 Peer attachment T1 income
M (SD)
Possible Range
α
7.94 (5.11) 7.00 (4.93) 2.43 (1.57) 1.93 (1.82) 1.19 (0.52) 0.21 (0.48) 1.76 (0.88) 3.77 (1.09) 7.32 (2.64)
1-40 1-40 1-5 1-8 1-8 0-4 0-4 1-5 1-10
.80 .78 NA NA .78 .85 .82 .69 NA
Note. NA = not applicable as the measure is based on one item. AUDIT = Alcohol Use Disorders Identification Test.
sport participation at T1 and perpetration of each violent behavior (physical IPV, emotional IPV, fighting, and assault) at T2, controlling for T1 violence and demographic variables (specifically, age, gender, education, and income). In Step 4, fully adjusted logistic regression models were performed to test relationships between sport participation at T1 and perpetration of each violent behavior at T2, controlling for T1 violence, demographics, and T1 common risk factors. In Step 5, two-way interactions were added to the fully adjusted models (Sport participation × Alcohol use, Gender × Alcohol use, and Gender × Sport participation). Finally, in Step 6 the three-way interaction (Gender × Alcohol use × Sport participation) was added to these models. Statistically significant interactions were probed using separate regression equations for sport participants and nonparticipants. For all effects, statistical significance was accepted as p < .05.
Results At T1, 60% (n = 1,347) of the sample participated in sport at least once in the past year. Of the sport participants, 23% participated 1 to 2 times per year, 13% participated 3 to 5 times per year, 9% participated 6 to 9 times per year, 11% participated 10 to 19 times per year, 8% participated 20 to 29 times per year, 5% participated 30 to 39 times per year, and 31% participated 40 or more times per year. Table 2 reports the rates of violent behaviors and AUDIT scores at each time point, as a function of sport participation (at least once in the past 12 months). Emotional IPV was the most prevalent behavior while assault was
Downloaded from jiv.sagepub.com at Australian National University on June 4, 2016
Scholes-Balog et al.
1511
the least prevalent. A statistically significantly greater proportion of nonsport participants reported perpetrating physical IPV (at least once in the past 12 months) at both T1 and T2, relative to sport participants. At T1, a statistically significantly greater proportion of nonsport participants also reported emotional IPV, relative to sport participants. There was no statistically significant difference in the proportion of sport participants and nonparticipants reporting fighting or assault at either time points. AUDIT scores were significantly higher for sport participants than for nonparticipants at both time points. The results of the partially adjusted logistic regression analyses are reported in Table 3. Each violent behavior at T1 was the strongest predictor of its respective violent behavior at T2. Sport participation was not a statistically significant predictor of any violent behavior at T2. Alcohol use problems at T1 predicted each violent behavior at T2. Females had significantly increased odds of perpetrating IPV (both emotional and physical) but significantly decreased odds of perpetrating assault or engaging in fights. Tables 4 to 7 present the fully adjusted logistic regression analyses for each violent behavior. As seen in Table 4, gender (female), T1 emotional IPV and T1 AUDIT were all associated with increased odds of T2 emotional IPV. However, frequency of sport participation at T1 did not predict T2 emotional IPV. There were also no statistically significant two-way or three-way interactions. Similarly, gender (female), T1 physical IPV and T1 cigarette use were all associated with increased odds of T2 physical IPV. There were no statistically significant two or three-way interactions (Table 5). As can be seen in Table 6, gender (female) was associated with reduced odds of T2 assault while T1 assault and antisocial peers were associated with increased odds of T2 assault. Again, sport participation at T1 did not predict T2 assault. There were also no statistically significant two-way or three-way interactions. Finally, as seen in Table 7, gender (female) was associated with reduced odds of T2 fighting while T1 alcohol use problems, T1 fighting, T1 education (low), antisocial peers, and cigarette use were all associated with increased odds of T2 fighting. Although there was no main effect of T1 sport participation on T2 fighting, there was a statistically significant interaction between sport participation and alcohol use problems (no other two-way or three-way interactions reached statistical significance). To probe this interaction, separate logistic regression analyses were completed as a function of sport participation. As can be seen in Table 8, for those who reported participating in sport at least once in the past 12 months, alcohol use problems at T1 were associated with an increased odds of fighting at T2, even when controlling for prior engagement in fights and common risk factors. However, for those who did not report participating in sport in the past 12 months, alcohol use
Downloaded from jiv.sagepub.com at Australian National University on June 4, 2016
1512
Downloaded from jiv.sagepub.com at Australian National University on June 4, 2016
% [95% CI]
Total
7.36 (5.53)
M (SD)
Nonparticipants
T1
7.94 (5.11)
M (SD)
Total
71.67 66.97 [68.05, 75.04] [64.65, 69.22] 21.79 [18.75, 16.59 [14.86, 25.16] 18.49] 9.94 [8.05, 11.83] 11.85 [10.55, 13.14] 1.55 1.79 [0.77, 2.34] [1.26, 2.32]
% [95% CI]
Nonparticipants
.0001
pb
6.36 (4.74)
M (SD)
M (SD) 7.40 (4.99)
Nonparticipants
Sport Participants
T2
67.90 [64.12, 71.47] 17.10 [14.33, 20.27] 10.05 [8.10, 12.01] 1.64 [0.82, 2.46]
% [95% CI]
% [95% CI]
66.10 [63.02, 69.05]
JIVXXX10.1177/0886260514567962Journal of Interpersonal ViolenceScholes-Balog et al.
Article
Relationships Between Sport Participation, Problem Alcohol Use, and Violence: A Longitudinal Study of Young Adults in Australia
Journal of Interpersonal Violence 2016, Vol. 31(8) 1501–1530 © The Author(s) 2015 Reprints and permissions: sagepub.com/journalsPermissions.nav DOI: 10.1177/0886260514567962 jiv.sagepub.com
Kirsty E. Scholes-Balog, PhD,1 Sheryl A. Hemphill, PhD,1,2,3,4 Peter J. Kremer, PhD,3 and John W. Toumbourou, PhD3,4
Abstract There is a growing body of evidence suggesting a link between sport participation and violent behavior outside of the sporting context. However, there have been few studies that have investigated the basis of this relationship. The current study examined longitudinal relationships between sport participation, problem alcohol use, and various violent behaviors, and whether sport participation moderates relationships between problem alcohol use and violence. The sample comprised 2,262 young adults (55% female, age range at Time 1 = 17-24 years) from Victoria, Australia, surveyed in 2010 and 2012. When controlling for common risk factors, substance use, and past violence, sport participation was not associated with any violent behaviors 2 years later. However, sport participation moderated the relationship between problem alcohol use and fighting, whereby problem
1Australian
Catholic University, Fitzroy, Victoria, Australia University of Melbourne, Parkville, Victoria, Australia 3Deakin University, Geelong, Victoria, Australia 4Murdoch Childrens Research Institute, Parkville, Victoria, Australia 2The
Corresponding Author: Kirsty E. Scholes-Balog, School of Psychology, Australian Catholic University, 115 Victoria Parade, Fitzroy, Victoria 3065, Australia. Email: [email protected]
Downloaded from jiv.sagepub.com at Australian National University on June 4, 2016
1502
Journal of Interpersonal Violence 31(8)
alcohol use was associated with engaging in fights 2 years later for sport participants, but not for nonparticipants. These findings suggest that it is not sport participation per se that influences later violence but the drinking norms or culture embedded within certain sporting contexts. Prevention approaches that address the drinking culture and social approval of excessive alcohol consumption within sporting contexts may reduce the incidence of violent behavior in the community. Keywords sport, alcohol use problems, violence, young adult Media reports of high profile athletes’ involvement in aggressive and violent incidents have focused attention on the relationship between sport participation and violence. While many of these violent or aggressive incidents take place on the field in the context of sports competition, there are also reports of violence against members of the community, and often with respect to women in intimate and nonintimate situations. Despite this anecdotal evidence of a link between sport involvement and community violence, there have been relatively few empirical studies that have been able to appropriately examine this relationship (Kimble, Russo, Bergman, & Galindo, 2010). Of the studies that do exist, many have found evidence to suggest that involvement in sport may be associated with violence and aggression outside of the sporting context. Participation in sport has been shown to be associated with family violence (Bloom & Smith, 1996), laboratory measures of aggression (D. B. Huang, Cherek, & Lane, 1999), sexual violence/aggression (Frinter & Rubinson, 1993; Koss & Gaines, 1993; Murnen & Kohlman, 2007), violent delinquency (Segrave, Moreau, & Hastad, 1985), and physical and/or verbal aggression/violence (Burton & Marshall, 2005; Endresen & Oleweus, 2005; Linville & Huebner, 2005; Nixon, 1997; O’Brien et al., 2012). Gender (Linville & Huebner, 2005; Miller, Melnick, Farrell, Sabo, & Barnes, 2006; Nixon, 1997; O’Brien et al., 2012), type of sport (e.g., contact or noncontact sport; Nixon, 1997), level of competitiveness (Bloom & Smith, 1996), and whether or not an individual perceives himself or herself as an athlete or “jock” (Miller et al., 2006) may also influence these relationships. It should be noted that not all studies have observed relationships between sport participation and violent behavior off-field (Kimble et al., 2010). For example, among college men, participation in sport (membership on a competitive sports team) was not found to be predictive of perpetration of verbal, physical, or sexual violence 3 months later when prior violent behavior was taken into account (Gidycz, Warkentin, & Orchowski, 2007).
Downloaded from jiv.sagepub.com at Australian National University on June 4, 2016
Scholes-Balog et al.
1503
There are a number of theories that help explain potential relationships between sport participation and violence. Social learning theory suggests that aggressive behaviors are learned through the confrontational nature of some sports (e.g., tackling, use of physical force, fighting skills in sports such as boxing or wrestling; Endresen & Oleweus, 2005). Furthermore, enactive learning and contact with aggressive role models during training and competition may enhance aggressive behaviors in the sporting context and the normalization or social approval of such behaviors (Endresen & Oleweus, 2005). Cultural spillover theory suggests that socially approved forms of aggression or violence can then have a carryover effect to more forbidden forms of aggression or violence, such as outside of the sporting context (Bloom & Smith, 1996). Participation in sport has also been consistently linked with alcohol use, with many studies showing that members of sporting clubs (Poortinga, 2007) and athletes (Gmel, Kuendig, & Daeppen, 2009; Leichliter, Meilman, Presley, & Cashin, 1998; O’Brien et al., 2012; O’Brien, Kolt, Webber, & Hunter, 2010) report greater alcohol consumption and more harm associated with alcohol consumption than nonsports people. For example, a large study of college students in the United States found that athletes reported more heavy episodic drinking, heavier alcohol use, and a greater level of drinking-related harms than nonathletes (Nelson & Wechsler, 2001). Furthermore, level of involvement in sport may influence the relationship between sport participation and alcohol use, with elite sports people reporting higher rates of hazardous drinking than both nonelite sports people and nonsports people (O’Brien, Ali, Cotter, O’Shea, & Stannard, 2007; O’Brien, Blackie, & Hunter, 2005; Sonderlund et al., 2014). There is also a well-established link between alcohol use and violent behavior. Experimental studies have shown that alcohol has an indirect causal influence on violent behavior, by way of cognitive, emotional, or physiological changes caused by alcohol consumption, which increase the likelihood of aggressive responses (for a review, see Exum, 2006). Longitudinally, alcohol use has also been found to predict later violent behavior (outside of the period of intoxication) in a number of studies (Dembo et al., 1993; Ellickson, Tucker, & Klein, 2003; Newcomb & McGee, 1989; Scholes-Balog, Hemphill, Kremer, & Toumbourou, 2013), while other studies have suggested bidirectional or reciprocal longitudinal relationships between alcohol use and violence (Scholes-Balog et al., 2013; White, Loeber, Stouthamer-Loeber, & Farrington, 1999; Windle, 1990; Zue, Zimmerman, & Cunningham, 2009). Longitudinal studies also support an indirect causal relationship between alcohol use and later violent behavior, with factors such as affiliation with antisocial and substance using peers appearing to explain these relationships (e.g., Scholes-Balog et al., 2013).
Downloaded from jiv.sagepub.com at Australian National University on June 4, 2016
1504
Journal of Interpersonal Violence 31(8)
While the bulk of the literature has examined violence in relation to sports participation and alcohol consumption separately, there is a small body of literature that has attempted to evaluate the relationship among all three factors (for a review, see Sonderlund et al., 2014). A qualitative study among athletes found that athletes often attribute alcohol use as the cause of their violent behavior outside of the sporting context (Pappas, McKenry, & Skilken Catlett, 2004). Similarly, off-field alcohol-related aggression and antisocial behavior was found to be common among professional Australian Football League players (Dietze, Fitzgerald, & Jenkinson, 2008). In this study, 26% of players surveyed reported getting involved in a fight (either verbal or physical) while drinking (Dietze et al., 2008). Likewise, a study of Australian university students found that university sport participants were more likely than nonsport participants to have displayed aggressive behavior (insulted or assaulted someone) when intoxicated in the past year (O’Brien et al., 2012). The university sport participant sample included participants in both individual (e.g., running, squash) and team sports (e.g., football, cricket). In this study, gender also played a role, with alcoholrelated violence and aggression more likely among male university sport participants than among female sport participants (O’Brien et al., 2012). Given the mounting evidence suggesting a higher rate of alcohol consumption and violence among athletes, relative to nonathletes, Sonderlund et al. (2014) suggested that sport participation may have a moderating effect on the positive relationship between alcohol and violence. However, this suggestion has yet to be empirically tested. Nonetheless, the context in which drinking takes place (and the associated dynamics of such context) has been shown to play a role in alcohol-related violence. For instance, public drinking (as compared with private drinking) is a strong predictor of violence (Rossow, 1996; Tomsen, 1997). Similarly, drinking among sporting spectators is thought to play a role in spectator aggression and violence at sporting matches (Bormann & Stone, 2001; Moore, Shepherd, Eden, & Sivarajasingam, 2007), with the social dynamics of the spectator crowd thought to also drive this relationship (Roberts & Benjamin, 2000). It should be noted that the bulk of the literature examining alcohol use, sport participation, and violence is cross-sectional in nature. One longitudinal study found that neither alcohol use nor sport participation predicted violence or aggression, when controlling for past history of violence and aggression (Gidycz et al., 2007). Another longitudinal study did not find sport participation itself to be predictive of violence, but rather “jock” identity (those who subjectively perceive themselves, and are perceived by others, as athletes or “jocks”) was associated with a higher frequency of violence (Miller et al., 2006). Furthermore, “jock” identity moderated the relationship between
Downloaded from jiv.sagepub.com at Australian National University on June 4, 2016
Scholes-Balog et al.
1505
alcohol use and violence, whereby alcohol use predicted violence only among nonjocks (Miller et al., 2006). There are a number of limitations of the published literature to date. First, the literature is dominated by studies of North American college samples (Sonderlund et al., 2014). Factors such as drinking cultures, sports team dynamics and subcultures, and social norms may vary between countries. For example, in Australia there is no named subculture synonymous with “jock,” though such a subculture may exist without name, or simply less pejoratively termed “sporty” or “athlete.” Nonetheless, literature investigating “jock” identity is not easily comparable in an Australian setting. Similarly, in countries where sport participation is not as valued or esteemed as in the United States, the customs associated with these sports may be less likely to influence participants’ behavior (Sonderlund et al., 2014). Furthermore, as mentioned previously, the vast majority of published studies investigating sport participation, alcohol use, and violence are of cross-sectional design and therefore are unable to assess temporal ordering or control for prior violent behavior when assessing relationships between sport participation and violence. In addition, most studies fail to control for other factors that may explain or account for any observed relationships between sport participation, alcohol use, and violence, such as use of illicit substances and affiliation with deviant peers (Herrenkohl et al., 2007; McMorris, Catalano, Kim, Toumbourou, & Hemphill, 2011; Scholes-Balog et al., 2013). Given the lack of research directed specifically at understanding the relationships between sport participation, alcohol use, and violence, together with the limitations outlined above, the present study intended to investigate longitudinal relationships between sport participation, problem alcohol use, and violence among a community-based Australian sample. Specifically, this study examines relationships with four violent behaviors: physical intimate partner violence (IPV), emotional IPV, fighting, and assault. It was hypothesized that (a) both alcohol use problems and frequency of sport participation will predict violent behavior 2 years later; (b) as suggested by Sonderlund et al. (2014), sport participation will moderate the relationship between alcohol use problems and violence; and (c) gender will moderate the relationship between alcohol use, sport participation, and violence.
Method Participants Young adults from Victoria, Australia, who were involved in the International Youth Development Study (IYDS) comprised the sample for this study. In
Downloaded from jiv.sagepub.com at Australian National University on June 4, 2016
1506
Journal of Interpersonal Violence 31(8)
2002, 3,948 Victorian students and their parents were approached to take part in this study. Of these, 2,927 parents provided informed consent (74%). Among those students who had parental informed consent, 2,884 (99%) were recruited into the IYDS and completed the survey. Original sampling for the IYDS has been described in detail elsewhere (McMorris, Hemphill, Toumbourou, Catalano, & Patton, 2007) and was designed to yield a staterepresentative sample of students in Grades 5, 7, and 9 (equivalent to average ages 11, 13, and 15 years). Briefly, the IYDS used a two-stage cluster sampling approach: (a) random selection of public and private schools in Victoria, Australia, stratified according to geographic location, using a probability proportionate to grade-level size sample procedure and (b) random selection of one class at each target grade level (Grades 5, 7, and 9 in 2002) within each school. Longitudinal surveying for the IYDS has occurred between the years 2002 and 2012. The sample for the current study comprised 2,262 (1,260 female, 1,000 male) young adults who were surveyed in both 2010 and 2012 as part of the IYDS longitudinal study and had data available for the measures of violence (this constituted 77% of the original sample who had entered the study in 2002). The current study utilizes these two time points as only questions related to the focus of this article were asked during these surveys. In 2010 (referred to as T1), participants ranged in age from 17 to 24 years of age (M = 21.00, SD = 1.67). In 2012 (referred to as T2), participants ranged in age from 19 to 26 years of age (M = 23.02, SD = 1.66).
Procedure Ethics approval was obtained from The University of Melbourne Human Ethics in Research Committee. Participants were contacted by mail, email, and/or phone and asked to complete the survey online after providing informed consent. Survey completion took approximately 60 min. After completion of the survey, participants received a small thank you gift (gift voucher).
Measures Survey items were drawn from a modified version of the Communities That Care youth survey (Glaser, Lee Van Horn, Arthur, Hawkins, & Catalano, 2005; Hemphill et al., 2011) that was adapted for use in young adults. Sport participation. Participation in sport was assessed at T1 with a question asking, “How many times in the past 12 months have you participated in organized sporting activities (organized by a club, association, or other type
Downloaded from jiv.sagepub.com at Australian National University on June 4, 2016
Scholes-Balog et al.
1507
of organization)?” This item was scored on an 8-point scale from never (1) to 40+ times (8). As this variable was severely skewed, and could not be rectified with transformation, it was dichotomized for analysis: sport participant (1) was defined as having engaged in sport at least once in the past 12 months, sport nonparticipant (0) was defined as not having engaged in sport at all in the past 12 months. Alcohol use problems. The Alcohol Use Disorders Identification Test (AUDIT; Babor, de la Fuente, Saunders, & Grant, 1992) was used to measure alcohol use problems at T1. The AUDIT consists of 10 items, each scored on a 5-point scale. Responses to each item are summed to obtain an overall score that ranges between 0 and 40. Higher scores are indicative of greater severity of alcohol-related problems, with scores more than 8 indicating a strong likelihood of harmful alcohol consumption (Babor et al., 1992). The AUDIT has been shown to be valid and reliable among college samples (e.g., Fleming, Barry, & Macdonald, 1991). Fighting. Engaging in fights was assessed at both T1 and T2. At T1, the item asked, “Have you ever gotten into physical fights with other people?” and the response options were “no,” “yes, but not in the past 12 months,” “yes, once in the past 12 months,” and “yes, more than once in the past 12 months.” At T2, the item asked, “How many times in the past year (12 months) have you got into physical fights with other people?” and the response options ranged from “never” to “10 or more times.” Given the different response options at each time point, for both T1 and T2, these items were dichotomized; a score of 1 indicated that the participant had engaged in fights at least once in the past 12 months, whereas a score of 0 indicated that the participant did not engage in fighting in the past 12 months. Assault. Perpetration of assault was assessed at both T1 and T2. At T1, the item asked, “Have you ever beat up someone so badly that they probably needed a doctor?” and had the same response options as the T1 fighting item. At T2, the item asked “How many times in the past year (12 months) have you beat up someone so badly that they probably needed to see a doctor?” and had the same response options as the T2 fighting item. As above, at T1 and T2 these items were dichotomized; a score of 1 indicated that the participant had perpetrated assault at least once in the past 12 months, whereas a score of 0 indicated that the participant had not perpetrated assault in the past 12 months. IPV. Perpetration of IPV in the past 12 months was assessed at both T1 and T2 with a scale containing five items, each assessing perpetration of different
Downloaded from jiv.sagepub.com at Australian National University on June 4, 2016
1508
Journal of Interpersonal Violence 31(8)
types of behaviors: (a) “I pushed, grabbed, slapped, shoved my partner, or threw something at my partner that could hurt”; (b) “I threatened to end our relationship or leave”; (c) “I threatened to hit or throw something at my partner”; (d) “My partner had a physical injury, sprain, bruise, or small cut because of a fight with me”; and (e) “I insulted, swore, or yelled at my partner.” The IPV questions were only asked of participants who indicated they had been/were currently in a relationship in the past 12 months (n = 1,545 at T1 and n = 1,567 at T2). These items were drawn from the revised Conflict Tactics Scale (Straus, Hamby, Boney-McCoy, & Sugarman, 1996), and in some cases, were modified for the purpose of this study. The frequency of each behavior was scored with response options “this has never happened to me” (0), “not in the past year, but it did happen before” (1), “once in the past year” (2), “twice in the past year” (3), “3-5 times in the past year” (4), “6-10 times in the past year” (5), “11-20 times in the past year” (6) and “more than 20 times in the past year” (7). Two measures of IPV were derived from these items: physical IPV (Items a, c, and d) and emotional IPV (Items b and e). To be consistent with the other measures of violence, these measures were dichotomized; a score of 1 indicated that the participant had perpetrated at least one IPV behavior at least once in the past 12 months, whereas a score of 0 indicated that the participant had not perpetrated IPV in the past 12 months. Common risk factors. A range of risk factors common to both alcohol use and violence were measured at T1 and controlled for in the analyses. Antisocial peers was assessed with a scale containing seven items asking about the number of peers that have engaged in certain antisocial behaviors in the past 12 months (e.g., “In the past year, how many of your best friends have stolen something worth more than $50AUD?”). Each item was scored on a 5-point scale, ranging from none of my friends (0) to 4 of my friends (4). Peer drug use was assessed with a scale containing seven items asking about the number of friends who had used certain types of drugs in the past 12 months (e.g., “In the past year, how many of your best friends have gotten drunk?”). This item was scored on the same rating scale as above. Finally, peer attachment was assessed with two items (e.g., “Think about your four best friends (the friends you feel closest to), in the past year, how many of your best friends have you shared your thoughts and feelings with?”). These items were rated on a 5-point scale from none of my friends (1) to 4 of my friends (5). For each of these scales, scale scores were obtained by taking the mean of the item responses. In addition, use of other substances were included as covariates in the analyses. Cigarette use was assessed with the item “Have you smoked cigarettes in the past year?” This item was scored on a 5-point scale from never
Downloaded from jiv.sagepub.com at Australian National University on June 4, 2016
Scholes-Balog et al.
1509
(1) to almost every day or every day (5). Cannabis use was assessed with the item “In the past year, on how many occasions, if any, have you used marijuana (pot, weed, grass)?” This item was scored on an 8-point scale ranging from never (1) to 40+ times (8). Finally, use of other illicit substances was assessed with eight items asking about the frequency of use of different types of illicit substances (lysergic acid diethylamide [LSD], cocaine, inhalants, stimulants, ecstasy, heroin, other illicit drugs, over the counter drugs for recreational/nonmedical purposes). Each item assessing use of each substance was rated on an 8-point scale ranging from never (1) to 40+ times (8). A composite measure of frequency of other illicit substance use was computed by taking the mean of the eight items. Finally, in addition to age and gender, two demographic variables were included as covariates in the regression analyses. Education was assessed with the question, “What was the highest year level at secondary school you completed?” This item was measured on a 5-point scale, ranging from year 8 or below (5) to year 12 or equivalent (1). Income was included as a proxy measure of socioeconomic status. Income was assessed with the question, “What is your usual, take-home, weekly income from all sources of support (this means after tax is taken out)?” This item was scored on a 10-point scale ranging from $0 (1) to $501 or more (10). Descriptive statistics for the measures can be found in Table 1.
Analyses Data were analyzed using the statistical software STATA IC for Windows (version 13; StataCorp, 2013). The analyses consisted of a number of steps. First, attrition was examined. Of those who were originally recruited into the IYDS in 2002, but were not included in the current study due to drop out, 63% (n = 381) were male. Those included in the current sample also had higher scores at the first wave of data collection (collected in 2002 when participants were aged 10-15 years) on measures of frequency of violence (beat someone up) and antisocial behavior (school suspension, stealing, arrest, carrying weapon, attacking someone) than those who did not complete both the young adult waves of the survey (collected in 2010 and 2012). However, there was no difference on measures of lifetime alcohol use measured in 2002 between those who completed the young adult waves and those who did not. Second, the variance inflation factor (VIF) of each independent variable in each analysis was examined for potential multicollinearity. All VIFs were less than 2.06, indicating no multicollinearity. In Step 3, partially adjusted logistic regression analyses were performed to test relationships between
Downloaded from jiv.sagepub.com at Australian National University on June 4, 2016
1510
Journal of Interpersonal Violence 31(8)
Table 1. Descriptive Statistics for Alcohol Use and Covariate Measures.
T1 AUDIT—Alcohol use problems T2 AUDIT—Alcohol use problems T1 Cigarette use T1 Cannabis use T1 Other illicit drug use T1 Antisocial peers T1 Friends’ drug use T1 Peer attachment T1 income
M (SD)
Possible Range
α
7.94 (5.11) 7.00 (4.93) 2.43 (1.57) 1.93 (1.82) 1.19 (0.52) 0.21 (0.48) 1.76 (0.88) 3.77 (1.09) 7.32 (2.64)
1-40 1-40 1-5 1-8 1-8 0-4 0-4 1-5 1-10
.80 .78 NA NA .78 .85 .82 .69 NA
Note. NA = not applicable as the measure is based on one item. AUDIT = Alcohol Use Disorders Identification Test.
sport participation at T1 and perpetration of each violent behavior (physical IPV, emotional IPV, fighting, and assault) at T2, controlling for T1 violence and demographic variables (specifically, age, gender, education, and income). In Step 4, fully adjusted logistic regression models were performed to test relationships between sport participation at T1 and perpetration of each violent behavior at T2, controlling for T1 violence, demographics, and T1 common risk factors. In Step 5, two-way interactions were added to the fully adjusted models (Sport participation × Alcohol use, Gender × Alcohol use, and Gender × Sport participation). Finally, in Step 6 the three-way interaction (Gender × Alcohol use × Sport participation) was added to these models. Statistically significant interactions were probed using separate regression equations for sport participants and nonparticipants. For all effects, statistical significance was accepted as p < .05.
Results At T1, 60% (n = 1,347) of the sample participated in sport at least once in the past year. Of the sport participants, 23% participated 1 to 2 times per year, 13% participated 3 to 5 times per year, 9% participated 6 to 9 times per year, 11% participated 10 to 19 times per year, 8% participated 20 to 29 times per year, 5% participated 30 to 39 times per year, and 31% participated 40 or more times per year. Table 2 reports the rates of violent behaviors and AUDIT scores at each time point, as a function of sport participation (at least once in the past 12 months). Emotional IPV was the most prevalent behavior while assault was
Downloaded from jiv.sagepub.com at Australian National University on June 4, 2016
Scholes-Balog et al.
1511
the least prevalent. A statistically significantly greater proportion of nonsport participants reported perpetrating physical IPV (at least once in the past 12 months) at both T1 and T2, relative to sport participants. At T1, a statistically significantly greater proportion of nonsport participants also reported emotional IPV, relative to sport participants. There was no statistically significant difference in the proportion of sport participants and nonparticipants reporting fighting or assault at either time points. AUDIT scores were significantly higher for sport participants than for nonparticipants at both time points. The results of the partially adjusted logistic regression analyses are reported in Table 3. Each violent behavior at T1 was the strongest predictor of its respective violent behavior at T2. Sport participation was not a statistically significant predictor of any violent behavior at T2. Alcohol use problems at T1 predicted each violent behavior at T2. Females had significantly increased odds of perpetrating IPV (both emotional and physical) but significantly decreased odds of perpetrating assault or engaging in fights. Tables 4 to 7 present the fully adjusted logistic regression analyses for each violent behavior. As seen in Table 4, gender (female), T1 emotional IPV and T1 AUDIT were all associated with increased odds of T2 emotional IPV. However, frequency of sport participation at T1 did not predict T2 emotional IPV. There were also no statistically significant two-way or three-way interactions. Similarly, gender (female), T1 physical IPV and T1 cigarette use were all associated with increased odds of T2 physical IPV. There were no statistically significant two or three-way interactions (Table 5). As can be seen in Table 6, gender (female) was associated with reduced odds of T2 assault while T1 assault and antisocial peers were associated with increased odds of T2 assault. Again, sport participation at T1 did not predict T2 assault. There were also no statistically significant two-way or three-way interactions. Finally, as seen in Table 7, gender (female) was associated with reduced odds of T2 fighting while T1 alcohol use problems, T1 fighting, T1 education (low), antisocial peers, and cigarette use were all associated with increased odds of T2 fighting. Although there was no main effect of T1 sport participation on T2 fighting, there was a statistically significant interaction between sport participation and alcohol use problems (no other two-way or three-way interactions reached statistical significance). To probe this interaction, separate logistic regression analyses were completed as a function of sport participation. As can be seen in Table 8, for those who reported participating in sport at least once in the past 12 months, alcohol use problems at T1 were associated with an increased odds of fighting at T2, even when controlling for prior engagement in fights and common risk factors. However, for those who did not report participating in sport in the past 12 months, alcohol use
Downloaded from jiv.sagepub.com at Australian National University on June 4, 2016
1512
Downloaded from jiv.sagepub.com at Australian National University on June 4, 2016
% [95% CI]
Total
7.36 (5.53)
M (SD)
Nonparticipants
T1
7.94 (5.11)
M (SD)
Total
71.67 66.97 [68.05, 75.04] [64.65, 69.22] 21.79 [18.75, 16.59 [14.86, 25.16] 18.49] 9.94 [8.05, 11.83] 11.85 [10.55, 13.14] 1.55 1.79 [0.77, 2.34] [1.26, 2.32]
% [95% CI]
Nonparticipants
.0001
pb
6.36 (4.74)
M (SD)
M (SD) 7.40 (4.99)
Nonparticipants
Sport Participants
T2
67.90 [64.12, 71.47] 17.10 [14.33, 20.27] 10.05 [8.10, 12.01] 1.64 [0.82, 2.46]
% [95% CI]
% [95% CI]
66.10 [63.02, 69.05]
